lable at ScienceDirect

Polymer xxx (2011) 1e7

Contents lists avai

Polymer

journal homepage: www.elsevier .com/locate/polymer

Fracture and fatigue response of a self-healing epoxy adhesive

Henghua Jin a,c,1, Gina M. Miller a,1, Nancy R. Sottos b,c, Scott R. White a,c,*

aAerospace Engineering, University of Illinois at Urbana-Champaign, USAbMaterials Science and Engineering, University of Illinois at Urbana-Champaign, USAcBeckman Institute, University of Illinois at Urbana-Champaign, USA

a r t i c l e i n f o

Article history:Received 30 November 2010Accepted 5 February 2011Available online xxx

Keywords:Self-healingAdhesivesEpoxy

* Corresponding author. Aerospace Engineering, UnChampaign, 104 S. Wright St., Urbana, IL, USA. Tel.: þ

E-mail address: swhite@illinois.edu (S.R. White).1 These authors contributed equally to this work.

0032-3861/$ e see front matter � 2011 Elsevier Ltd.doi:10.1016/j.polymer.2011.02.011

Please cite this article in press as: Jin H, etj.polymer.2011.02.011

a b s t r a c t

A self-healing epoxy adhesive for bonding steel substrates is demonstrated using encapsulated dicy-clopentadiene (DCPD) monomer and bis(tricyclohexylphosphine)benzylidine ruthenium (IV) dichloride(Grubbs’ first generation) catalyst particles dispersed in a thin epoxy matrix. Both quasi-static fractureand fatigue performance are evaluated using the width-tapered-double-cantilever-beam specimengeometry. Recovery of 56% of the original fracture toughness under quasi-static fracture conditionsoccurs after 24 h healing at room temperature conditions. Complete crack arrest is demonstrated forfatigue test conditions that render neat resin and control samples failed. Inspection of fracture surfacesby electron microscopy reveals evidence of polymerized DCPD after healing. These results are the firstmechanical assessment of self-healing for thin (ca. 360 mm) films typical of adhesives applications.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Inspired by living systems, self-healing polymers are designedto repair damage whenever and wherever it occurs. Severaldifferent conceptual approaches to self-healing have been exploredover the past several years. In the first, microcapsules containingreactive chemical species are incorporated into the native polymermatrix and upon crack damage release their contents and undergoa healing reaction [1e4]. In the second, a vascular network isembedded and serves as a reservoir for healing agent(s) forsequestration and distribution throughout the polymer matrix[5e7]. The third approach utilizes inherently reversible bonding inthe matrix polymer to affect healing via thermally reversiblereactions [8e10] or reformation of hydrogen bonds [11].

Although a variety of self-healing polymers have been reported[12], application and performance as adhesives for structuralbonding remains unexplored. Thin epoxy structural adhesive filmsare commonly used in aerospace and automotive industries forbonding metallic and composite substrates. They are also the crit-ical performance factor for composite laminate repairs ofaluminum aerospace structures [13]. During the repair process,a composite patch (doubler) is attached to the damaged aluminumskin using an epoxy adhesive film. Failure of the repair is most

iversity of Illinois at Urbana-1 2173331077.

All rights reserved.

al., Fracture and fatigue resp

commonly adhesive debonding and can be difficult to detect usingnondestructive techniques. Composite double repair has also beenrecently adapted to steel structures [14] in which a boron-epoxycomposite doubler patch is co-cured with an epoxy based adhesivefilm to form a bonded repair.

Self-healing of epoxy was initially achieved by incorporatinga microencapsulated low molecular weight monomer (dicyclo-pentadiene, DCPD) and a solid phase chemical catalyst (bis(tricy-clohexylphosphine)benzylidine ruthenium (IV) dichloride: Grubbs’first generation catalyst) within an epoxy matrix. Epoxy containing10 wt.% microcapsules (ca. 200 mm) containing DCPD and 2.5 wt.%Grubbs’ catalyst yielded as much as 75% recovery of virgin fracturetoughness [1]. By systematically studying size and concentration,healing efficiencies as high as 90% were achieved [2]. Improve-ments in catalyst protection schemes [15] and morphology [16] ledto reductions in catalyst concentration and improvements in thehealing kinetics.

Translating these results to adhesives presents a number ofunique technical challenges. Adhesives are typically applied as thinfilms over large planar areas and as such, place geometricconstraints on the maximum size of microcapsules that can beused. The presence of a microencapsulated liquid phase healingagent and a solid phase catalyst must not interfere with adhesion ofthe substrate surface and the native epoxy matrix. Finally, thehealing chemistry must provide not only good adhesion to theepoxy matrix, but also to the substrate material(s).

In this paper we demonstrate a self-healing epoxy adhesivesuitable for bonding steel substrates. The materials system is

onse of a self-healing epoxy adhesive, Polymer (2011), doi:10.1016/

Fig. 1. Scanning electron microscope images of (a) as-received Grubbs’ first generation catalyst, (b) DCPD-filled microcapsules with poly(urea-formaldehyde) shell wall.

H. Jin et al. / Polymer xxx (2011) 1e72

adapted from our previous studies of bulk epoxy healing consistingof microencapsulated DCPD monomer and Grubbs’ first generationcatalyst [1]. The quasi-static fracture and fatigue response ofadhesively bonded steel samples are presented for a variety ofmaterial formulations. Self-healing epoxy adhesives have potentialto increase the service life of bonded structures and of laminaterepairs for both aerospace and infrastructure applications.

2. Materials

The adhesive matrix consisted of EPON� 828 epoxy resin(DGEBA) (Miller-Stephenson), cured with 12 pph Ancamine�

diethylenetriamine (DETA) (Air Products). Bothmaterials were usedas-received. The healing chemistry consisted of bis(tricyclohex-ylphosphine)benzylidine ruthenium (IV) dichloride (Grubbs’ first

Fig. 2. WTDCB specimen. (a) Geometry of WTDCB specimen consisting of adhesively bondecrack. Aluminum foil is used to control adhesive thickness. Dimensions are given in the unitwith Grubbs’ catalyst and DCPD microcapsules.

Please cite this article in press as: Jin H, et al., Fracture and fatigue respj.polymer.2011.02.011

generation) catalyst (SigmaeAldrich) used as-received (Fig.1a), andendo-dicyclopentadiene (Alfa Aesar), which was distilled prior toencapsulation. Dibutyl phthalate (DBP) (SigmaeAldrich) wasencapsulated as-received and incorporated in control specimens.Silane coupling agent (3-glycidyloxypropyl)trimethoxysilane (Sig-maeAldrich) was used to improve the adhesion between steelsubstrates and the epoxy adhesive. A 25 mm thick fluoropolymerrelease film (A4000R) (Airtech International) was molded in spec-imen to serve as a starter crack. The substrate was A36 structuralsteel (Speedy Metals).

The DCPD healing agent was stabilized with 150 ppmp-tert-butylcatechol (Acros Organics) and encapsulated followingestablished techniques [17] by in situ polymerization of urea-formaldehyde. Microcapsules with average diameter of 130 (�23)mm were produced at 550 RPM agitation rate with a shell wall of

d A36 steel adherends. A 25 mm thick fluoropolymer release ply is used to create a pre-s of mm. (b) Optical microscopy of cross section of a self-healing adhesive incorporated

onse of a self-healing epoxy adhesive, Polymer (2011), doi:10.1016/

Table 1Summary specimen types and adhesive components.

Test Specimen Adhesive

Matrix Catalyst (wt.%) Microcapsule (wt.%)

Fracture Reference control 828/DETA e e

Self-activated control 828/DETA 2.5 e

Self-healing 828/DETA 2.5 15 (DCPD)Fatigue Neat 828/DETA e e

Control 828/DETA 2.5 15 (DBP)Self-healing 828/DETA 2.5 15 (DCPD)

Fig. 3. Representative virgin and healed load versus crack opening displacement (COD)curves for reference specimens. The triangles represent the values used to calculate themean critical loads of both virgin and healed tests. The upper dash line is the meanvirgin critical load (228 N) and the lower dash line is the mean healed critical load(115 N).

H. Jin et al. / Polymer xxx (2011) 1e7 3

approximately 160e220 nm thickness [Fig. 1b]. Mechanicalrupture of the shell wall triggers in situ polymerization of DCPD byring-opening-metathesis-polymerization (ROMP) with Grubbs’catalyst producing a tough, highly crosslinked polymer [18]. DBPwas encapsulated by the same encapsulation procedure at550 RPMyielding capsuleswith average diameter of 122 (�23) mm.Unreactive with Grubbs’ catalyst, DBP provided a control casewhere the effects of the healing reaction were isolated from othermaterial effects.

3. Experimental methods

3.1. Quasi-static fracture testing

The fracture toughness of adhesively bonded steel substrates ismeasured by preparing and testing width-tapered-double-canti-levered-beam (WTDCB) specimens (Fig. 2a). The WTDCB geometryprovides a crack length independent measurement of Mode-Ifracture toughness [19e21], critical to accurate analysis of healingperformance. Assuming the adhesive layer is thin and does notsignificantly contribute to the total specimen thickness, linearelastic fracture analysis of the WTDCB geometry yields the mode-Istress intensity factor [22],

KI ¼ 2Pk

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi3Ea�

1� n2�Esh3s

s(1)

where P is the applied load, k is the taper ratio (for this study, k ¼ 3is used), vs (0.33) is the adherend (substrate) Poisson’s ratio, hs isthe substrate thickness, and Ea (3.4 GPa [23]) and Es (200 GPa) arethe modulus of elasticity of the adhesive and the substrate,respectively. The healing efficiency is defined as the ratio of thehealed fracture toughness to the virgin fracture toughness [2]. Forthe WTDCB geometry the healing efficiency reduces to the ratio ofcritical fracture loads of the healed and virgin tests,

h ¼ KhealedIC

KvirginIC

¼ Phealedc

Pvirginc

: (2)

Fracture toughness testing was performed using an Instron8500 load frame with a 4000 N load cell in displacement control ata rate of 20 mm/min until a crack opening displacement (COD,measured by crosshead extension) of d ¼ 2 mm was reached. Forsome control specimens, healing solutions were injected directlyinto the crack plane while the specimen was held open under load.After the initial (virgin) fracture test, the specimens were unloadedto allow the fracture surfaces to come into contact and heal. After24 h of healing at room temperature, the specimens were reloadeduntil a total COD of 2 mm was reached.

3.2. Fatigue testing

The fatigue performance of the self-healing adhesive was alsoinvestigated using the WTDCB test geometry. Specimens were

Please cite this article in press as: Jin H, et al., Fracture and fatigue respj.polymer.2011.02.011

tested in the Instron load frame while applying a cyclic stress topropagate the crack. A 2 Hz haversine waveform was applied witha maximum stress intensity of Kmax

I ¼ 0:42 MPaffiffiffiffiffim

pand a stress

intensity ratio (R ¼ KminI =Kmax

I ) of 0.1. Crack growthwasmonitoredby the measured specimen compliance [22],

Da ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiCEsh3s12k

s� a0 (3)

where C ¼ d=P is the compliance of the specimen and a0 is pre-crack length. Fatigue specimens were fractured after healing andthe final crack lengths were measured optically to confirm thecalculated final crack length.

3.3. Specimen fabrication

Steel adherends were prepared for bonding by manual sandingusing 80 grit sandpaper, cleaning with compressed air and acetone,followed by wiping using acetone to remove debris from thesurface. Adherends were then rinsed with a 1 vol.% silane couplingagent solution, dried at room temperature for 30 min, and cured ina 60 �C oven for 1 h. Six layers of 25-mm aluminum foil were thenattached to both the beginning of tapered region and the end of thesample (see Fig. 2b) to control the thickness of adhesive layer.

Epoxy adhesive was prepared by mixing EPON 828 with 12 pphDETA curing agent. The epoxy mixture was degassed for 15 min tothoroughly remove entrapped air bubbles before pouring onto eachprepared adherend surface. For samples that contained capsulesand/or catalyst, these were mixed into the epoxy resin at theappropriate concentration and then degassed for several additionalminutes. Thereafter, the mixture was evenly spread across theadherend surfaces and a 25 mm thick fluoropolymer release ply wasplaced at the beginning of the taper region extending approxi-mately 5 mm along the length of the specimen to serve as a pre-crack. Two adherends were mated together and the specimen wascured at room temperature for 24 h followed by 24 h at 35 �C. Afterthe cure cycle was complete, two loading blocks with through holesfor pin loading were attached to the end of each specimen byscrews (Fig. 2c).

Three types of specimens were prepared for the quasi-staticfracture study (see Table 1). The reference control specimens

onse of a self-healing epoxy adhesive, Polymer (2011), doi:10.1016/

Table 2Summary of quasi-static fracture test results.

Specimen # of Samples Fracture Toughness (MPa�m1/2) Healing Efficiency (%) Adhesive Thickness (mm)

Virgin Healed

Reference 7 0.65 � 0.04 0.26 � 0.07 41 � 11 363 � 21Self-activated 8 0.78 � 0.08 0.33 � 0.07 44 � 13 359 � 23Self-healing 12 0.82 � 0.08 0.45 � 0.19 56 � 24 375 � 21

H. Jin et al. / Polymer xxx (2011) 1e74

contained only neat EPON 828/DETA epoxy adhesive. Healing forthe reference specimen occurs via injection of a pre-catalyzedmixture of 0.5 mL DCPD and 2.5 mg Grubbs’ catalyst into the crackplane using a syringe. A second set of controls (self-activated)containing 2.5 wt.% Grubbs’ catalyst in the epoxy adhesive werealso prepared. These controls provided evidence of catalyst survivaland activity in the epoxy matrix. Healing for the self-activatedsamples was accomplished by injection of 0.5 mL DCPD monomerinto the crack plane. Fully in situ specimens contained both 2.5 wt.%Grubbs’ catalyst and 15 wt.% DCPD-filled microcapsules in theEPON 828/DETA epoxy adhesive (Fig. 2b).

Three types of fatigue specimens were prepared for testingfollowing the same procedures established for fracture toughnesssamples. In addition to neat epoxy adhesive and fully in situ self-healing adhesive, control samples weremade incorporating 15wt.%DBP-filled microcapsules and 2.5 wt.% Grubbs’ catalyst in order toisolate the effect of self-healing on the extension of fatigue life(Table 1).

3.4. Fractography

Fracture surface morphology of both quasi-static and fatiguefracture samples were examined using a field emission environ-mental scanning electron microscope (Philips XL30 ESEM-FEG).Samples were sputter-coated with gold-palladium before imaging.

4. Results and discussion

4.1. Quasi-static fracture

A representative loading curve for a reference specimen isshown in Fig. 3 for the initial (virgin) test and the healed test. In the

Fig. 4. Representative virgin and healed load versus crack opening displacement (COD)curves for self-activated specimens. The triangles represent the values used for themean critical loads of both virgin and healed tests. The upper dash line is the meanvirgin critical load (271 N) and the lower dash line is the mean healed critical load(138 N).

Please cite this article in press as: Jin H, et al., Fracture and fatigue respj.polymer.2011.02.011

virgin case load increases linearly with COD until a critical load(w260 N) after which stick-slip, unstable crack propagation occursat a consistent critical load until reaching a total COD of 2 mm. Atthis point, a pre-catalyzed mixture of 0.5 mL DCPD monomer and2.5 mg Grubbs’ catalyst is injected into the crack plane, and thespecimen unloaded. After 24 h of healing at room temperature, thespecimen is retested and follows the virgin loading curve untilreaching a healed critical load (w125 N) at which point the crackadvances through the healed region. Once the crack fully propa-gates through the initial fracture region, the loading curve followsthe same contour as the virgin unloading path until a total COD of2 mm is reached. The average virgin and healed critical loads arecalculated based on the individual propagation events during thevirgin and healed tests, (as indicated in Fig. 3). For this particularspecimen, the average virgin and healed critical loads are 228 N and115 N, respectively. The healing efficiency as defined in Eq. (2) ish ¼ 51%. The average healing efficiency for all reference tests ish¼ 41�11% and the full results for all quasi-static fracture tests arepresented in Table 2.

A representative test result for a self-activated test specimen ispresented in Fig. 4. Again, loading is initially linear up to a criticalload for crack propagation in both virgin and healed tests. Theaverage virgin and healed critical loads for this specimen are 271 Nand 138 N, respectively, yielding a healing efficiency of h¼ 51%. Theaverage healing efficiency over eight tests for self-activated speci-mens is h ¼ 44 � 13% (Table 2). There is a significant increase invirgin toughness for self-activated specimens compared to the neatresin adhesive [23].

Fig. 5 shows typical loading curves for the fully in situ self-healing specimen. In this case the virgin loading curve possesses thesame characteristic stick-slip propagation mechanism until reach-ing a COD of 2 mm. However, the healed loading curve is distinctly

Fig. 5. Representative virgin and healed load versus crack opening displacement (COD)curves for self-healing specimens. The triangles represent the values used for the meancritical load of virgin test. The healed critical load is the mean value of the databetween circles. The upper dash line is the mean virgin critical load (281 N) and thelower dash line is the mean healed critical load (151 N).

onse of a self-healing epoxy adhesive, Polymer (2011), doi:10.1016/

Fig. 6. Quasi-static fracture plane after healing of a self-healing specimen. (a) Optical images of mating fracture surfaces showing cohesive fracture. (b) SEM image of healed fracturesurface showing poly(DCPD) film (blue) formed during self-healing. (c) SEM image of fracture surface at the end of the sample created during imaging (unhealed) revealing cracktails indicative of toughening mechanism. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

H. Jin et al. / Polymer xxx (2011) 1e7 5

unique from reference and self-activated test cases. After initiatingpropagation of the healed crack, stable and continuous crackpropagation occurs throughout the healed region. This character-istic behavior was observed for all twelve specimens tested. The

Fig. 7. Representative fatigue response of neat, control and self-healing fatigue spec-imens. Samples were loaded using 2 Hz haversine waveform at a maximum stressintensity of Kmax

I ¼ 0:42 MPaffiffiffiffiffim

pand a stress intensity ratio of R ¼ 0.1.

Please cite this article in press as: Jin H, et al., Fracture and fatigue respj.polymer.2011.02.011

average virgin critical load (281 N) is obtained by sampling thecritical loads associated with the propagation events in the virginloading trace. For the healed test, we obtain the mean value duringcrack propagation as designated by the symbols on the loadingtrace. The initial point is defined based on a 10% deviating in slopefrom the virgin test and the final point corresponds to the peak loadduring propagation of the healed crack. Sampling over this CODrange yields the average healed critical load of 151 N which corre-sponds to a healing efficiency of h ¼ 54%. The average healing effi-ciency for all in situ samples tested is h ¼ 56 � 24% (Table 2). Again,the virgin toughness is significantly improved from the neat resinadhesive case [23].

Cohesive failure through the self-healing adhesive was observedfor all test cases, an essential requirement for the rupture ofembedded microcapsules and activation of the healing mechanism.Surface preparation of the steel adherends by mechanical abrasion,cleaning, and silane coupling agent was applied to promote cohesive

Table 3Summary of fatigue test results.

Specimen # of Samples Crack Growth Rate (mm/cycle) Cycles to Failure

Neat 4 2.72 � 1.91 14,940e61,980Control 4 0.26 � 0.06 278,220e465,060Self-healing 4 0a Na

a Crack arrest occurs after 150,000 cycles in all cases.

onse of a self-healing epoxy adhesive, Polymer (2011), doi:10.1016/

Fig. 8. SEM images of fatigue fracture surfaces for (a) neat, (b) control, and (c) self-healing specimens showing poly(DCPD) wedge (blue). (For interpretation of the references tocolor in this figure legend, the reader is referred to the web version of this article.)

H. Jin et al. / Polymer xxx (2011) 1e76

failure (Fig. 6a). Incorporation ofmicrocapsules and catalyst particlesinto the adhesive also promotes crack path deflection from thesubstrate/adhesive interface into the adhesive layer for the self-healing and self-activated control specimens [24e26]. As such, bothtypes of specimens exhibited cohesive failure. However, referencecontrol specimens exhibited a mixture of both cohesive and adhe-sive failure possibly a result of a change in residual stresses and/orlocalized T-stress within the epoxy adhesive bondline [27,28].

The improvement in virgin fracture toughness for the adhesivewith the addition of self-healing components is consistent withestablished toughening mechanisms for bulk self-healing polymers[23]. The fracture plane of a virgin test for a self-healing specimenshows extensive crack tail formation characteristic of a crackpinning toughening mechanism (Fig. 6c).

Fractography of healed fracture surfaces reveal the existence ofa crosslinked polymer film on the fracture surface (Fig. 6b) indic-ative of the in situ formation of crosslinked poly(DCPD) from thehealing reaction. Importantly, this autonomic reaction leads to therecovery of 56% of the virgin fracture toughness on average.

4.2. Fatigue behavior

The representative fatigue response of neat, control and self-healing specimens are shown in Fig. 7. All neat epoxy adhesivespecimens failed quickly for the given test conditions (<62,000cycles), as shown in Table 3. Control specimens containing Grubbs’catalyst and microcapsules containing an unreactive core (DBP)demonstrated significant increase in fatigue life and a reduction ininitial crack growth rate (see Table 3). Similar effects have beenpreviously demonstrated in bulk self-healing polymers due to theactivation of toughening mechanisms with the addition of micro-capsules and as a result of hydrodynamic crack-tip shielding afterthe release of the encapsulated fluid into the crack plane [29,30].

In stark contrast to both the neat epoxy adhesive and controlspecimens, full crack arrest occurred for all in situ self-healingspecimens tested (Fig. 7). After a brief period of initial crack growth(ca. 10 mm) the crack growth rate rapidly decreases until completearrest occurs. Crack arrest for self-healing specimens occurs largelydue to the in situ formation of a polymer wedge (as shown inFig. 8c) in the crack plane preventing full unloading of the crack tip,and a reduction in the effective stress intensity range [29,30].Additionally, the polymerized healing agent can promote adhesivebonding of the crack faces and a further reduction of the effectivestress intensity range. Eventually, the effective stress intensityrange is reduced sufficiently and no further crack growth occurs.

Examination of the neat epoxy fatigue specimen by SEM revealsrelatively smooth and striated morphology as shown in Fig. 8a. Bycontrast, the fracture plane for control and self-healing fatiguespecimens reveal many out-of-plane fracture features and remnants

Please cite this article in press as: Jin H, et al., Fracture and fatigue respj.polymer.2011.02.011

of microcapsules (Fig. 8b and c). For self-healing specimens, poly(DCPD) wedge features are also present on the fracture plane.

5. Conclusions

A self-healing epoxy adhesive has been demonstrated by incor-porating a two-part self-healing system of 15 wt.% micro-encapsulated dicyclopentadiene (DCPD) monomer and 2.5 wt.%Grubbs’ first generation catalyst particles. The addition of bothcomponents to the neat resin epoxy increased the virgin fracturetoughness by 26% and scanning electron microscopy of fracturesurfaces reveals evidence of toughening through crack pinning.Recovery of quasi-static mode-I fracture toughness was assessedusing width-tapered-double-cantilever-beam (WTDCB) test speci-mens. Self-healing specimens recovered 56% of the virgin fracturetoughness on average after 24 h healing at room temperature. Thefatigue response of self-healing specimens was also investigated ata maximum stress intensity factor of 0.42 MPa�m1/2 and a stressintensity ratioof 0.1. Theneatepoxyadhesive specimens failedwithin62,000 cycles under these conditions while all self-healing samplesexhibited complete crack arrest.

Acknowledgments

The authors gratefully acknowledge funding support fromNational Science Foundation (Grant # CMS 05-27965) and SandiaNational Laboratories (SPO 378467) with Dr. Dennis Roach asProgramManager. The authorswould also like to thankKent Elam inthe Aerospace Engineering Machine Shop for help manufacturingthe WTDCB adherends. Testing was completed at the AdvancedMaterials Testing and Engineering Lab, with assistance of PeterKurath, Gavin Horn and Rick Rottet. Electron microscopy was per-formed in the Imaging Technology Group, Beckman Institute forAdvanced Science and Technology at the University of Illinois, withthe assistance of Scott Robinson.

References

[1] White SR, Sottos NR, Moore JS, Geubelle PH, Kessler M, Brown E, et al. Nature2001;409(6822):794e7.

[2] Brown EN, Sottos NR, White SR. Experimental Mechanics 2002;42(4):372e9.[3] Caruso MM, Blaiszik BJ, White SR, Sottos NR, Moore JS. Advanced Functional

Materials 2008;18:1898e904.[4] Xiao DS, Yuan YC, Rong MZ, Zhang MQ. Polymer 2009;50(13):2697e775.[5] Williams GJ, Trask RS, Bond IP. Composites A 2007;38(6):1525e32.[6] Toohey KS, Sottos NR, Lewis JA, Moore JS, White SR. Nature Materials

2007;6:581e5.[7] Toohey KS, Sottos NR, White SR. Experimental Mechanics 2009;49:707e17.[8] Chen XX, Dam MA, Ono K, Mal A, Shen HB, Nutt SR, et al. Science 2002;295

(5560):1698e770.[9] Paisted TA, Nemat-Nasser S. Acta Materialia 2007;55(17):5684e96.

onse of a self-healing epoxy adhesive, Polymer (2011), doi:10.1016/

H. Jin et al. / Polymer xxx (2011) 1e7 7

[10] Hayes SA, Zhang W, Branthwaite M, Jones FR. Journal of the Royal SocietyInterface 2007;4(13):381e7.

[11] Cordier P, Tournilhac F, Soulie-Ziakovic C, Leibler L. Nature 2008;451(7181):977e80.

[12] Blaiszik BJ, Kramer SLB, Olugebefola SC, Moore JS, Sottos NR, White SR. AnnualReview of Materials Research 2010;40:179e211.

[13] Baker AA, Rose LRF, Jones R. Advances in the bonded composite repair ofmetallic aircraft structure, vol. 1. Elsevier; 2002. pp. 485e516.

[14] Roach D, Rackow K. Sandia report. SAND2005e3195; 2005.[15] Rule JD, Brown EN, Sottos NR, White SR. Advanced Materials 2005;17

(2):205e8.[16] Jones AS, Rule JD, Moore JS, Sottos NR, White SR. Journal of the Royal Society

Interface 2007;4(13):395e403.[17] Brown EN, Kessler MR, Sottos NR, White SR. Journal of Microencapsulation

2003;20(6):719e30.[18] Kessler MR, White SR. Journal of Polymer Science: Part A: Polymer Chemistry

2002;40(14):2373e83.[19] Kessler MR, Sottos NR, White SR. Composites Part A: Applied Science and

Manufacturing 2003;34(8):743e53.

Please cite this article in press as: Jin H, et al., Fracture and fatigue respj.polymer.2011.02.011

[20] Hwang W, Han KS. Journal of Composite Materials 1989;23(12):396e430.[21] Jyoti A, Gibson RF, Newza GM. Composite Science and Technology 2005;65

(1):9e18.[22] Miller G. Self-healing adhesive film for composite laminate repairs on metallic

structures, MS thesis. University of Illinois at Urbana Champaign; 2007.[23] Brown EN, White SR, Sottos NR. Journal of Material Science 2004;39(5):

1703e10.[24] Grujicic M, Sellappan V, Omar MA, Seyr N, Obieglo A, Erdmann M, et al.

Journal of Material Processing Technology 2008;197(1e3):363e73.[25] Faber KT, Evans AG. Acta Metallurgica 1983;31(4):565e76.[26] Faber KT, Evans AG. Acta Metallurgica 1983;31(4):577e84.[27] Fleck NA, Hutchinson JW, Suo Z. International Journal of Solids Structures

1991;27(13):1683e703.[28] Chen B, Dillard DA. International Journal of Adhesion and Adhesive 2001;21

(5):357e68.[29] Brown EN, White SR, Sottos NR. Composites Science and Technology 2005;65

(15e16):2466e73.[30] Brown EN, White SR, Sottos NR. Composites Science and Technology 2005;65

(15e16):2474e80.

onse of a self-healing epoxy adhesive, Polymer (2011), doi:10.1016/